Persona at invocation

Before conversational narration, read Persona: and Style: in the active project's ## Agent skills block from CLAUDE.md or AGENTS.md. Prefer the file containing the block, then an existing file; ties use CLAUDE.md. A symlink pair is one file. Read the saved value anew on each invocation, including from a subdirectory using the project root. No accessible project or no line means ordinary behavior. Do not search another project or global settings for this preference.

During the Persona at invocation stage, archmage on either line loads this skill's own references/archmage.md for the active workflow. off or an absent value leaves ordinary behavior active. An unknown value leaves ordinary behavior active and gets a brief explanation when conversational output is allowed; it does not stop the work. Explicit conversation instructions override the saved voice without writing settings. A request to enable Archmage for this workflow also loads the local reference.

Apply the voice only to lead-agent conversation. Deliverables, specialist roles, reply-only responses, and JSON-only output retain their contracts, with no added narration. End the persona with this workflow unless the user requests otherwise or a Style: line names archmage, which keeps the voice on for the whole session.

Vision

Keep one evidence-based roadmap that explains milestone priorities and the next useful action.

Honor explicit user instructions and decisions already made over workflow defaults. Continue through the requested result; ask only about unresolved choices that would materially change it. Reporting progress alone does not authorize implementation work. If a real permission or missing prerequisite prevents completion, explain it and finish independent authorized work.

Shared contracts

• The saved pointer or roadmap label identifies the roadmap; legacy body markers remain readable. Maps and build efforts link up through a Milestone: line in their own bodies; closing them requires no roadmap write.
• Reconciliation derives activity from members and completion from outcome evidence, preserving explicit Confirmed done: overrides. Snapshot the body and guard updates against concurrent changes. Incomplete reads leave the saved roadmap unchanged.
• Default reporting explains progress and remaining gaps without inventing dates or estimates. Counts support the assessment. Use milestone names and linked issue titles in user-facing output.

<SKILL_DIR> is the absolute directory containing this file. Substitute it directly in bundled script commands.

Route

1. Resolve: load references/tracker.md to find the tracker and roadmap. For a local roadmap, also load references/scratch.md.
2. Chart: when no roadmap exists, load references/chart.md. It loads the body schema and milestone guidance. Create the roadmap and save its pointer, then reconcile existing work and report.
3. Reconcile: for an existing roadmap, load references/reconcile.md, including for add <name>, done <n>, reopen <n>, or order <n n ...>. It loads the schema, applies the edit, and refreshes status under the body guard.
4. Report: load references/report.md after charting or reconciliation. A creation or update is complete when authorized writes and the next-step report are delivered. Draft and read-only requests end with the proposed or derived assessment without writes. Continue a broader explicitly authorized request after that result.

you-pick, "make the decisions", or "you pick" authorizes recommended choices for charting. Preserve decisions already supplied without requiring that token.

Bundled readers and writers

Use scripts/tickets.sh for tracker operations with the resolved adapter flags. It supports GitHub with git and gh, and Linear or Jira with Python's standard library. Exit 3 means the token cannot write; the charting reference describes the local fallback.

Use scripts/map.sh for GitHub map frontier reads. Its optional repository argument is owner/repo, not the ticket script's adapter flags. Other trackers use the Wayfinding operations in their tracker configuration.

The optional persona reference loads only during the persona stage above. Workflow-specific references load only at their named stage.

References